Maksim Dromov
Offer Maksim work on your next project.
Private message 
Kyiv,
Ukraine
16 hours 5 minutes back
available for hire
1 proposal made
age 33 years
on the service 7 years
Rating
Successful projects
Average rating
Language proficiency level
Українська: fluent
Русский: fluent
English: intermediate
Skills and abilities
Programming
Services
Photo, Audio & Video
Writing
Portfolio
-
250 USD Creation of realistic photographs with the client's product
AI & Machine Learning
Client's task:
The owner of an online men's suit store wanted to showcase products on real models - without a photo studio, shooting costs, and manual labor. A solution was needed that works quickly and does not require technical knowledge from employees.
Solution:
… I developed an automated system based on n8n with a Telegram interface that transforms product photos into ready-made photos of models in suits.
The user simply:
- Sends a product photo
- Adds a desired background
- Chooses a model from 4 options
The system automatically parses the page, uploads images (if the product has a link), analyzes the suit, and sends everything to Gemini. In about 1 minute, 3 realistic photos of the model in the suit are ready, saved in Google Drive.
Result for the business:
The client completely eliminated the photo studio from regular expenses. New products appear in the catalog on the day of arrival, not a week later. The system operates without the involvement of a developer - any employee can manage it in 3 clicks.
Additionally, high-quality visuals in the product card directly influence the purchasing decision: the buyer sees the suit on a real person, not on a mannequin or hanger, which reduces doubts and increases conversion.
Numbers:
The cost of 1 set of 3 photos is approximately $0.75-1 compared to $150–300 for studio shooting.
Budget savings on content - up to 87%
Generation speed - 20-30 seconds per photo
Launching a new product in the catalog - in 2 minutes instead of 5–7 days
Stack:
n8n | Telegram Bot API | GPT-4o Mini | Gemini 3.1-flash-image | Google Drive |
n8n DataTable (storing user data and query history)
-
175 USD AI fitting room for the buyer — selection of products during purchases
AI & Machine Learning
Client's task:
The owner of a men's clothing store regularly goes on buying trips. At the location, it is necessary to quickly understand: will a specific item match the existing suits in the assortment?
Solution:
… I developed an automated system based on n8n with a Telegram interface. The buyer takes a photo of the item in the store and sends the picture. The system automatically uploads 5 pre-loaded models in suits from Google Drive and generates a hint with clear fitting rules for each pair. The system automatically determines the type of item and applies the necessary rules. Gemini generates 5 photos and returns the result directly in Telegram — the buyer sees all combinations and makes a purchasing decision on the spot.
Result for the business
Purchases have become more accurate: the buyer makes decisions not based on intuition, but on visual results. Items that do not match the assortment are filtered out in the store — before money is spent. This reduces the percentage of unsuccessful purchases and increases the average turnover of goods in the catalog.
Numbers
5 photos with different suits per request
Time from sending the photo to the result ~1-2 minutes
Cost of one check ~0.9-1.2$
Stack
n8n · Telegram Bot API · Gemini 3.1 Flash Image · Google Drive · JavaScript Code Nodes
-
150 USD Telegram bot for learning foreign words with a repetition system
Bot Development
Client's task:
A personal tool was needed for learning German words, which operates on the principle of spaced repetition, where words are repeated not consecutively, but depending on how well you know them.
Solution:
… I developed a Telegram bot based on n8n and Supabase. The user sends a German word - the bot automatically translates it via GPT and saves it with the status "New." Then, a system with three testing modes operates with a smart change of statuses: a correct answer advances the word, while an incorrect one returns it for review. After each answer, regardless of the result, GPT generates 3 example sentences with this word and its breakdown - the word is reinforced through context, not rote memorization. New → Semi-new → Learned. Words with the status "Learned" are no longer shown, and mistakes are highlighted in a separate mode and processed separately.
Result for the business:
The user focuses only on those words that truly need attention, does not waste time on what they already know, and does not miss those moments where mistakes are made. The system automatically adapts to each user. The bot is easily customizable for any language - just change the prompt.
Numbers:
- 3 testing modes: new words, mistakes, repetitions
- 4 word statuses: New → Semi-new → Learned / Mistake
- Customizable for any language without changing the architecture
Stack:
n8n · Telegram Bot API · GPT-4o Mini · Supabase · JavaScript Code Nodes
-
150 USD Telegram bot with characters that have individual memory
Bot Development
Client's task:
The agency needed a tool to automate communication with the audience via Telegram - with several unique characters, each having its own personality, dialogue style, and remembering the chat history with a specific user. The goal is to create an engaging communication experience that retains the audience and scales without hiring people.
Solution:
… Developed a Telegram bot based on n8n with 4 AI characters. Each character has a unique personality, communication style, and system prompt. The user chooses whom they want to communicate with, after which the bot maintains a natural conversation on behalf of the selected character. The system stores the last 100 messages for each "user - character" pair.
Business result:
The agency received a scalable audience engagement tool that operates 24/7 without the involvement of live operators. Each user receives a personalized communication experience - the bot remembers them, responds in the character of the chosen persona, and maintains interest. This directly impacts the time spent in the bot and audience return. The agency completely eliminates the cost of managers. The bot makes no mistakes, responds with the same quality at 3 AM on a Sunday as it does during a working noon. Zero human factor, zero downtime, and as a result, more engaged users, higher retention, and higher overall profit.
Figures:
- 4 unique characters with individual personalities and prompts
- 100 context messages for each "user - character" pair
- Switching between characters - with one click
- Operates 24/7 without operators
Stack:
n8n · Telegram Bot API · GPT-4o Mini · n8n Memory Buffer · n8n DataTable
-
350 USD AI stylist with online fitting for an online store
AI & Machine Learning
Client's task:
The owner of a men's suit store wanted to give customers the ability to see themselves in a suit before purchasing — directly on the website, without a fitting room and without visiting the store. The selection should be personalized: taking into account body type, skin tone, and the occasion for which the suit is needed.
Solution
… An automated system was developed based on n8n. The customer uploads their photo and selects an event. GPT-4o Mini analyzes body type, skin tone, and face shape, after which GPT-4.1 selects 5 most suitable suits from the catalog, along with a shirt, tie, and shoes for each. Gemini generates fitting photos: the customer sees themselves in each of the 5 looks. Direct links to product pages are automatically added to each photo, allowing immediate purchase.
Result for the business:
The store receives a tool that works like a personal stylist 24/7, without salary and human factors. The customer sees themselves in the suit before payment, receives a complete look tailored to their occasion and body type, and can immediately purchase everything with one click. This significantly increases conversion: a person comes for a suit but leaves with a complete look.
Numbers:
- 5 personalized looks with fitting in one request
- 4 direct links to products (suit, shirt, tie, shoes)
- 4 selection scenarios
- Time from photo to result ~1 minute
- Cost of one session ~1-1.5$
Stack
n8n · GPT-4o Mini (appearance analysis) · GPT-4.1 (suit and accessory selection) · Gemini 3.1 Flash Image (fitting visualization generation) · Google Sheets (product catalog) · Supabase (user data storage) · HTML Parser
-
150 USD Automatic parser of competitor reviews for online stores
AI & Machine Learning
Client's task:
The owner of an online store wanted to know which products sell best among competitors. Customer reviews are a direct indicator of demand: the more reviews there are and the fresher they are, the more actively the product sells. A tool was needed that collects this data automatically across all stores simultaneously.
Solution:
… I developed an automated system based on n8n that parses all products and all reviews from a specified list of stores. The list is stored in Google Sheets, and the system automatically visits each site, determines which review system is used, and collects data through the corresponding API. The results are exported to a separate Google Sheets table, breaking down reviews by month. For each product, not only the total number is visible, but also the dynamics.
Result for the business:
The client sees which products are currently selling the most actively and directs the budget towards purchasing these items. Instead of spending weeks on manual monitoring of competitor stores, they receive ready-made analytics with one click and make decisions based on data rather than assumptions.
Figures:
- 10 competitor stores monitored
- ~75,000 products and ~190,000 reviews collected in one run
- Support for 4 review systems: Yotpo (v1 and v3), Reviews.io, Judge.me, Okendo
- Full data collection cycle - without human involvement
Stack:
n8n · Google Sheets · Yotpo API (v1 + v3) · Reviews.io API · Judge.me API · Okendo API · JavaScript Code Nodes · HTTP request
-
100 USD AI agency website for visual content for e-commerce
AI Art
Task:
A professional showcase website was needed to present AI services for generating visual content for e-commerce brands — convincingly, without unnecessary text, and with real examples of results. The site was meant to encourage potential clients to get in touch, rather than just inform.
Solution:
… I created the website entirely without the involvement of a developer or designer. I developed the structure, content, and logic of the page using Claude, and the design and interface using Lovable. The site is built around a specific outcome: a section with demo before/after results in five niches, an interactive chat layout shows what the working process looks like from the inside, and a block with business advantages addresses objections to the call.
Result:
A fully functional agency website built from scratch - without a developer, without a designer, without an agency. The cost of creation — just time. The site overcomes the first barrier of trust: potential clients see real examples, understand the process, and know how to get in touch.
Numbers:
- 5 niches with demo results before/after
- $0 on developer and designer
- Launch time - 3 days
Tools:
Claude (structure, content, code) · Lovable (design and UI)
-
400 USD AI content automation for clothing — macro photography, styling
AI & Machine Learning
Client's task:
The owner of the online store wanted not just quality product photos - he needed complete visual content: macro shots of each item, structured descriptions for the catalog, and live photographs of models in outfits made from his assortment. Plus - the ability to quickly try on any new item on a ready photo without re-shooting.
Solution:
… I created an automated pipeline on n8n consisting of three sequential modules, each of which performs its task and passes the result further.
How it works:
Part 1 - Macro photography of the product:
The system scans a folder on Google Drive and retrieves uploaded photos. Gemini generates 8 professional macro shots with a clean white studio background - overall view, details of the cut, fabric texture. Each product receives a named folder on Google Drive.
Part 2 - Analysis and description.
The first photo from each folder is taken, and GPT-4o mini conducts a detailed analysis: category, color and color group, formality, subcategory, cut, combinations with other items. All this is stored in Google Sheets and becomes the basis for the third part.
Part 3 - Model photo and item replacement.
Through a Telegram bot, the client selects a model and uploads the item they want to try on to the ready outfit. GPT-4o mini-stylist forms the look: a top with the highest potential is selected from the catalog, then a suitable bottom is chosen according to the color compatibility and formality algorithm. Gemini generates a photo of the model in this look in a smart casual style. After that, the item uploaded by the client is automatically overlaid on the ready photo - the system recognizes its type (outerwear, top, or bottom) and replaces the corresponding element of the outfit, keeping the model's face, background, and pose unchanged.
Result for the business:
The client receives a complete visual package for each item in the catalog - from product photography to the look on the model - without a studio, photographer, or stylist. Any new item can be "tried on" to an already ready photo in seconds. The content scales without increasing costs.
Numbers:
- 1 original photo → 8 macro shots + description + image on the model
- 4 types of models to choose from
- 3 categories of clothing with unique prompts for each
- 2 options for the final shot (item unbuttoned / buttoned)
- 0 manual actions after uploading
Stack:
n8n · GPT-4o mini (OpenAI) · Gemini 3.1 Flash (Google AI) · Google Drive API · Google Sheets · Telegram Bot API · JavaScript
-
125 USD AI script generator for video content about store products
AI & Machine Learning
Client's task:
The owner of an online store needs a YouTube channel, Instagram, and TikTok to promote their products. Problem: writing each script takes a lot of time, and the quality depends on mood and inspiration. A tool is needed that generates ready-made scripts of high quality, tailored to a specific product and a specific platform.
Solution:
… An automated system based on n8n with a Telegram interface has been developed. The user inputs the topic, sends a photo of the product, and selects the platform - YouTube or Instagram/TikTok. GPT-4o Mini analyzes the product. Then GPT-5.2 generates a complete script structured as "fact → twist → path → solution → meaning." For YouTube, this is an extended script lasting 5-7 minutes; for short videos, a dynamic script lasting 30-90 seconds with three hook options. The finished script is automatically saved in Google Docs, and the link is sent via Telegram.
Result for the business:
The client consistently receives free traffic and audience trust, which converts into sales without advertising costs due to regular video content. The client has stopped wasting time writing scripts manually; the system transforms each product in the catalog into a potential video — and does this in 40 seconds.
Numbers:
- 2 script formats: YouTube (5-7 min) and short video (30-90 sec)
- 3 hook options for short videos — to choose from
- Script generation time ~30-40 seconds
Stack
n8n · Telegram Bot API · GPT-4o Mini (product analysis) · GPT-5.2 (script generation) · Google Docs · n8n DataTable
-
300 USD AI stylist for an online store. 10 looks for one product
AI & Machine Learning
Client's task:
The owner of an online store wanted to solve the main question of the buyer: "what should I wear this with?" An instrument was needed that automatically showcases a suit in various looks — without a stylist, without a photo studio, and without manual work.
Solution:
… I developed an automated system based on n8n, which generates 10 ready-made photos of a model in a suit from a single product link. The employee selects a model from 4 options and a style. The system automatically parses the page, analyzes the suit, generates 9 recommendations for looks based on combination logic, and creates photos for each look. A base photo on a neutral background plus 9 variations with different shirts, ties, shoes, and accessories, etc.
Result for the business:
Each suit receives a ready mini lookbook with 10 photos. The content immediately goes into the product card, social media, and advertising campaigns. The buyer sees specific looks tailored to their style - this eliminates doubts at the selection stage, increases the time spent on the product page, and directly converts into sales. Work that previously required a stylist, photographer, and several days now takes 2 minutes and costs $2.
Numbers:
- 10 ready photos per request — base + 9 looks
- Cost of the complete set ~$2-2.5 versus thousands $$$ for a professional lookbook
- Speed ~2-3 minutes for the entire package
- 2 styles to choose from: Formal and Smart Casual
Stack:
n8n · Telegram Bot API · GPT-4o Mini · GPT-4.1 · Gemini 3.1 Flash Image · Google Drive · n8n DataTable (storing sessions, images, and analytics for each user)
Activity
| Latest proposals 10 | Budget | Added | Deadlines | Proposal | |
|---|---|---|---|---|---|
|
Development of a Telegram bot for AI generation of explicit content using the model's appearance.
203 USD
|
|||||
Traffic arbitration in crypto tg
45 USD
|
|||||
|
Ladies
11 USD
|
|||||
|
Реклама канала
56 USD
|
|||||
|
Настройка таргетированной рекламы в Европе и США
68 USD
|
|||||
|
Настройка рекламы у фейсбук
45 USD
|
|||||
|
Таргетолог Instagram для фитнесс-блога
63 USD
|
|||||
|
Запуск и продвижение платного авторского Телеграм канала
68 USD
|
|||||
|
Ищу таргетолога в В2В
113 USD
|
|||||
|
Таргетированная реклама
124 USD
|
